61
60
# and add in all file versions
63
@deprecated_function(zero_eight)
65
64
def greedy_fetch(to_branch, from_branch, revision=None, pb=None):
65
"""Legacy API, please see branch.fetch(from_branch, last_revision, pb)."""
66
66
f = Fetcher(to_branch, from_branch, revision, pb)
67
67
return f.count_copied, f.failed_revisions
71
class Fetcher(object):
72
"""Pull revisions and texts from one branch to another.
74
This doesn't update the destination's history; that can be done
75
separately if desired.
78
If set, pull only up to this revision_id.
82
last_revision -- if last_revision
83
is given it will be that, otherwise the last revision of
72
class RepoFetcher(object):
73
"""Pull revisions and texts from one repository to another.
76
if set, try to limit to the data this revision references.
86
79
count_copied -- number of revisions copied
88
count_weaves -- number of file weaves copied
81
This should not be used directory, its essential a object to encapsulate
82
the logic in InterRepository.fetch().
90
def __init__(self, to_branch, from_branch, last_revision=None, pb=None):
91
if to_branch == from_branch:
92
raise Exception("can't fetch from a branch to itself")
93
self.to_branch = to_branch
94
self.to_weaves = to_branch.weave_store
95
self.to_control = to_branch.control_weaves
96
self.from_branch = from_branch
97
self.from_weaves = from_branch.weave_store
98
self.from_control = from_branch.control_weaves
84
def __init__(self, to_repository, from_repository, last_revision=None, pb=None):
99
86
self.failed_revisions = []
100
87
self.count_copied = 0
88
if to_repository.control_files._transport.base == from_repository.control_files._transport.base:
89
# check that last_revision is in 'from' and then return a no-operation.
90
if last_revision not in (None, NULL_REVISION):
91
from_repository.get_revision(last_revision)
93
self.to_repository = to_repository
94
self.from_repository = from_repository
95
# must not mutate self._last_revision as its potentially a shared instance
96
self._last_revision = last_revision
98
self.pb = bzrlib.ui.ui_factory.nested_progress_bar()
99
self.nested_pb = self.pb
102
self.nested_pb = None
103
self.from_repository.lock_read()
105
self.to_repository.lock_write()
109
if self.nested_pb is not None:
110
self.nested_pb.finished()
111
self.to_repository.unlock()
113
self.from_repository.unlock()
116
"""Primary worker function.
118
This initialises all the needed variables, and then fetches the
119
requested revisions, finally clearing the progress bar.
121
self.to_weaves = self.to_repository.weave_store
122
self.to_control = self.to_repository.control_weaves
123
self.from_weaves = self.from_repository.weave_store
124
self.from_control = self.from_repository.control_weaves
101
125
self.count_total = 0
102
self.count_weaves = 0
103
self.copied_file_ids = set()
104
126
self.file_ids_names = {}
106
self.pb = bzrlib.ui.ui_factory.progress_bar()
109
self.from_branch.lock_read()
127
pp = ProgressPhase('Fetch phase', 4, self.pb)
111
self._fetch_revisions(last_revision)
130
revs = self._revids_to_fetch()
134
self._fetch_weave_texts(revs)
136
self._fetch_inventory_weave(revs)
138
self._fetch_revision_texts(revs)
139
self.count_copied += len(revs)
113
self.from_branch.unlock()
116
def _fetch_revisions(self, last_revision):
117
self.last_revision = self._find_last_revision(last_revision)
118
mutter('fetch up to rev {%s}', self.last_revision)
119
if (self.last_revision is not None and
120
self.to_branch.has_revision(self.last_revision)):
123
revs_to_fetch = self._compare_ancestries()
125
raise InstallFailed([self.last_revision])
126
self._copy_revisions(revs_to_fetch)
127
self.new_ancestry = revs_to_fetch
129
def _find_last_revision(self, last_revision):
130
"""Find the limiting source revision.
132
Every ancestor of that revision will be merged across.
134
Returns the revision_id, or returns None if there's no history
135
in the source branch."""
138
self.pb.update('get source history')
139
from_history = self.from_branch.revision_history()
140
self.pb.update('get destination history')
142
return from_history[-1]
144
return None # no history in the source branch
143
def _revids_to_fetch(self):
144
mutter('fetch up to rev {%s}', self._last_revision)
145
if self._last_revision is NULL_REVISION:
146
# explicit limit of no revisions needed
148
if (self._last_revision != None and
149
self.to_repository.has_revision(self._last_revision)):
147
def _compare_ancestries(self):
148
"""Get a list of revisions that must be copied.
150
That is, every revision that's in the ancestry of the source
151
branch and not in the destination branch."""
152
self.pb.update('get source ancestry')
153
self.from_ancestry = self.from_branch.get_ancestry(self.last_revision)
155
dest_last_rev = self.to_branch.last_revision()
156
self.pb.update('get destination ancestry')
158
dest_ancestry = self.to_branch.get_ancestry(dest_last_rev)
161
ss = set(dest_ancestry)
163
for rev_id in self.from_ancestry:
165
to_fetch.append(rev_id)
166
mutter('need to get revision {%s}', rev_id)
167
mutter('need to get %d revisions in total', len(to_fetch))
168
self.count_total = len(to_fetch)
171
def _copy_revisions(self, revs_to_fetch):
173
for rev_id in revs_to_fetch:
177
if self.to_branch.has_revision(rev_id):
179
self.pb.update('copy revision', i, self.count_total)
180
self._copy_one_revision(rev_id)
181
self.count_copied += 1
184
def _copy_one_revision(self, rev_id):
185
"""Copy revision and everything referenced by it."""
186
mutter('copying revision {%s}', rev_id)
187
rev_xml = self.from_branch.get_revision_xml(rev_id)
188
inv_xml = self.from_branch.get_inventory_xml(rev_id)
189
rev = serializer_v5.read_revision_from_string(rev_xml)
190
inv = serializer_v5.read_inventory_from_string(inv_xml)
191
assert rev.revision_id == rev_id
192
assert rev.inventory_sha1 == sha_string(inv_xml)
193
mutter(' commiter %s, %d parents',
196
self._copy_new_texts(rev_id, inv)
197
parents = rev.parent_ids
198
new_parents = copy(parents)
199
for parent in parents:
200
if not self.to_branch.has_revision(parent):
201
new_parents.pop(new_parents.index(parent))
202
self._copy_inventory(rev_id, inv_xml, new_parents)
203
self.to_branch.revision_store.add(StringIO(rev_xml), rev_id)
204
mutter('copied revision %s', rev_id)
206
def _copy_inventory(self, rev_id, inv_xml, parent_ids):
207
self.to_control.add_text('inventory', rev_id,
208
split_lines(inv_xml), parent_ids,
209
self.to_branch.get_transaction())
211
def _copy_new_texts(self, rev_id, inv):
212
"""Copy any new texts occuring in this revision."""
213
# TODO: Rather than writing out weaves every time, hold them
214
# in memory until everything's done? But this way is nicer
215
# if it's interrupted.
216
for path, ie in inv.iter_entries():
217
self._copy_one_weave(rev_id, ie.file_id, ie.revision)
219
def _copy_one_weave(self, rev_id, file_id, text_revision):
220
"""Copy one file weave, esuring the result contains text_revision."""
221
# check if the revision is already there
222
if file_id in self.file_ids_names.keys( ) and \
223
text_revision in self.file_ids_names[file_id]:
225
to_weave = self.to_weaves.get_weave_or_empty(file_id,
226
self.to_branch.get_transaction())
227
if not file_id in self.file_ids_names.keys( ):
228
self.file_ids_names[file_id] = to_weave.names( )
229
if text_revision in to_weave:
231
from_weave = self.from_weaves.get_weave(file_id,
232
self.from_branch.get_transaction())
233
if text_revision not in from_weave:
234
raise MissingText(self.from_branch, text_revision, file_id)
235
mutter('copy file {%s} modified in {%s}', file_id, rev_id)
237
if to_weave.numversions() > 0:
238
# destination has contents, must merge
153
return self.to_repository.missing_revision_ids(self.from_repository,
155
except errors.NoSuchRevision:
156
raise InstallFailed([self._last_revision])
158
def _fetch_weave_texts(self, revs):
159
texts_pb = bzrlib.ui.ui_factory.nested_progress_bar()
161
file_ids = self.from_repository.fileids_altered_by_revision_ids(revs)
163
num_file_ids = len(file_ids)
164
for file_id, required_versions in file_ids.items():
165
texts_pb.update("fetch texts", count, num_file_ids)
167
to_weave = self.to_weaves.get_weave_or_empty(file_id,
168
self.to_repository.get_transaction())
169
from_weave = self.from_weaves.get_weave(file_id,
170
self.from_repository.get_transaction())
171
# we fetch all the texts, because texts do
172
# not reference anything, and its cheap enough
173
to_weave.join(from_weave, version_ids=required_versions)
174
# we don't need *all* of this data anymore, but we dont know
175
# what we do. This cache clearing will result in a new read
176
# of the knit data when we do the checkout, but probably we
177
# want to emit the needed data on the fly rather than at the
179
# the from weave should know not to cache data being joined,
180
# but its ok to ask it to clear.
181
from_weave.clear_cache()
182
to_weave.clear_cache()
186
def _fetch_inventory_weave(self, revs):
187
pb = bzrlib.ui.ui_factory.nested_progress_bar()
189
pb.update("fetch inventory", 0, 2)
190
to_weave = self.to_control.get_weave('inventory',
191
self.to_repository.get_transaction())
193
child_pb = bzrlib.ui.ui_factory.nested_progress_bar()
240
to_weave.join(from_weave)
241
except errors.WeaveParentMismatch:
242
to_weave.reweave(from_weave)
244
# destination is empty, just replace it
245
to_weave = from_weave.copy( )
246
self.to_weaves.put_weave(file_id, to_weave,
247
self.to_branch.get_transaction())
248
self.count_weaves += 1
249
self.copied_file_ids.add(file_id)
250
self.file_ids_names[file_id] = to_weave.names()
251
mutter('copied file {%s}', file_id)
195
# just merge, this is optimisable and its means we don't
196
# copy unreferenced data such as not-needed inventories.
197
pb.update("fetch inventory", 1, 3)
198
from_weave = self.from_repository.get_inventory_weave()
199
pb.update("fetch inventory", 2, 3)
200
# we fetch only the referenced inventories because we do not
201
# know for unselected inventories whether all their required
202
# texts are present in the other repository - it could be
204
to_weave.join(from_weave, pb=child_pb, msg='merge inventory',
212
class GenericRepoFetcher(RepoFetcher):
213
"""This is a generic repo to repo fetcher.
215
This makes minimal assumptions about repo layout and contents.
216
It triggers a reconciliation after fetching to ensure integrity.
219
def _fetch_revision_texts(self, revs):
220
"""Fetch revision object texts"""
221
rev_pb = bzrlib.ui.ui_factory.nested_progress_bar()
223
to_txn = self.to_transaction = self.to_repository.get_transaction()
226
to_store = self.to_repository._revision_store
228
pb = bzrlib.ui.ui_factory.nested_progress_bar()
230
pb.update('copying revisions', count, total)
232
sig_text = self.from_repository.get_signature_text(rev)
233
to_store.add_revision_signature_text(rev, sig_text, to_txn)
234
except errors.NoSuchRevision:
237
to_store.add_revision(self.from_repository.get_revision(rev),
242
# fixup inventory if needed:
243
# this is expensive because we have no inverse index to current ghosts.
244
# but on local disk its a few seconds and sftp push is already insane.
246
# FIXME: repository should inform if this is needed.
247
self.to_repository.reconcile()
252
class KnitRepoFetcher(RepoFetcher):
253
"""This is a knit format repository specific fetcher.
255
This differs from the GenericRepoFetcher by not doing a
256
reconciliation after copying, and using knit joining to
260
def _fetch_revision_texts(self, revs):
261
# may need to be a InterRevisionStore call here.
262
from_transaction = self.from_repository.get_transaction()
263
to_transaction = self.to_repository.get_transaction()
264
to_sf = self.to_repository._revision_store.get_signature_file(
266
from_sf = self.from_repository._revision_store.get_signature_file(
268
to_sf.join(from_sf, version_ids=revs, ignore_missing=True)
269
to_rf = self.to_repository._revision_store.get_revision_file(
271
from_rf = self.from_repository._revision_store.get_revision_file(
273
to_rf.join(from_rf, version_ids=revs)
276
class Fetcher(object):
277
"""Backwards compatibility glue for branch.fetch()."""
279
@deprecated_method(zero_eight)
280
def __init__(self, to_branch, from_branch, last_revision=None, pb=None):
281
"""Please see branch.fetch()."""
282
to_branch.fetch(from_branch, last_revision, pb)
added
removed
bzrlib/fetch.py
